Identifying pathogenic processes by integrating microarray data with prior knowledge

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Identifying molecular pathways in disease etiology is crucial but challenging.
• High-throughput methods often yield gene sets with poor overlap to known disease pathways.
• Understanding molecular processes underlying disease progression requires novel analytical approaches.

Purpose of the Study:

• To present a novel Bayesian statistical method for identifying functional molecular groups from differentially expressed genes.
• To improve the interpretation of genome-wide screen findings in disease research.
• To uncover novel connections and functional modules involved in disease etiology.

Main Methods:

• Utilizing Bayesian statistics to group co-regulated genes from microarray data.
• Incorporating external molecular interaction data as priors for group assignments.
• Employing Markov chain Monte Carlo (MCMC) sampling for reliable grouping.

Main Results:

• Simulations demonstrated improved group identification accuracy compared to traditional clustering, particularly with small sample sizes.
• Application to a heart failure dataset identified clusters related to extracellular matrix and carbohydrate metabolism.
• Analysis of a melanoma dataset revealed a main cluster associated with keratinocyte differentiation.

Conclusions:

• The developed method successfully identified clusters overlapping with known pathogenic processes.
• The approach also highlighted novel molecular connections extending beyond classical disease pathways.
• This facilitates a deeper understanding of molecular mechanisms in disease.